All numbers are not created equally, and it is important to understand the type of measurement system data values fall into. Typically, data is organized into one of four measurement systems: nominal, ordinal, interval, and ratio.

 

Nominal values distinguish one occurrence from another. Each value in this system represents something unique where there is no relationship between numbers. For example, a land use value of 4 – agricultural land – is not half the land use of 8, which is coniferous forest. Land use categories, zip codes, and telephone numbers are examples of nominal values.

 

 

 

 

Ordinal values describe placement or position, such as athletes on a podium. However, ordinal values cannot answer “how much more” or other questions regarding magnitude. A person who achieves the highest score on a test did not necessarily do twice as well as the person who received the 2nd highest score. Rankings, such as the agricultural value of soil is a good example of ordinal values.

 

 

Interval values fall on a linear scale but cannot be related to a true zero point. The measurement between intervals is always equal in this measurement system, such as the amount of time in an hour. Examples include pH, time of day, temperature.

 

 

 

 

The ratio measurement system is based off linear scales with a fixed zero point. Length, area, volume, age, and density are all examples of ratio values. All ratio values can be used in mathematical equations to yield meaningful results. Distance to the nearest road provides a helpful example. A location 2 kilometers from a road is twice as far as a location 1 kilometer from that road.

 

 

Adding land use categories (nominal), agricultural value of soils (ordinal), soil pH (interval), and distance from roads (ratio) results in a surface of meaningless values. A vale of 20 may denote a cell with a land use value of 5, soil with an agricultural viability rank of 6, and a soil pH of 8 that is 1 kilometer away from the nearest road.

 

Simply combining these values leaves out any sense of preference for the criteria. By transforming data onto a common preference scale, a phenomenon’s overall preference can be captured.

 

 

Transforming data onto a common scale allows a meaningful combination of data within a suitability model.

 

Each of the following layers of have been transformed onto a scale ranging from 1 to 10, with 1 denoting least suitable for farming and 10 indicating the most preferable areas.

 

• Land use
• Transformed land use

 

• Soil agricultural value
• Transformed agricultural value

 

• Soil pH
• Transformed soil pH

 

• Distance from roads
• Transformed distance from roads

Preference scales are either ratio or interval measurement systems . Common scales range from 1 to 10, 1 to 9, 0 to 1, and 1 to 100. Different criteria can now be combined together since they are on the same preference scale.

 

It is recommended that the preference must be consistent within a criteria, for example, a 5 preference is half as preferred as a 10 preference. It is also recommended that the preferences between criteria be consistent – a 5 preference on the distance from roads surface provides the same preference as a 5 of the resulting transformed land use criteria.

 

A 1 to 10 preference scale was used in the model below which contains 4 criteria. A location assigned a 40 would have the most preferred distance from roads, the best land use, ideal pH, and most valuable agricultural soil.

 

Click below to compare suitability surfaces.

 

• Suitability surface without transformed values
• Suitability surface with transformed values

 

 

 

Acknowledgements

We thank Steven Lamonde of Johnson State College and the Vermont Center for Geographic Information for their contributions.